Assessments of functional outcome and usage of prosthesis during activities of daily living (ADL) of lower limb amputees has gained increasing importance to support evidence-based practice (e.g., issue of underprescription and overprescription of prosthetic components). One of the most critical end products of these assessments is to scale amputees from the least to the most functional. Current practice is to use the US Medicare Functional Classification Levels, including four mobility grades or K-levels.1 Clinicians can classify their patients using a wide range of instruments, which can be used separately or in combination, as recommended in recent guidelines.2,3 An overview of the resources and comprehensiveness of the output of instruments that are most commonly used is presented in Figure 1.
SURROGATE MEASUREMENTS OF FUNCTIONAL OUTCOME
According to Parker et al.,4 the capacity to undertake ADL can be defined as “a participant's ability to walk and move with his or her prosthesis and ambulation aids (canes, crutches, or walkers) in a standardized environment (rehabilitation and clinics).” In most cases, this capacity is assessed after or during the fitting of the prosthesis using standardized instruments based on the following:
- Self-reports (e.g., Amputee Activity Survey, Prosthetic Profile of the Amputee and Locomotor Capabilities Index,5 Russek's code, Prosthetic Evaluation Questionnaire and Orthotic Prosthetic Users,6 Questionnaire for Persons with a Transfemoral Amputation,7 Special Interest Group in Amputee Medicine (SIGAM)8),
- Physical tasks (e.g., 2-minute walk, 6-minute walk, functional ambulation profile, timed get-up and go, and amputee mobility predictor with prosthesis).3,9 The performance is expressed in unit of time or distance.
Both types of instruments are easy to administer in clinical settings and require little resources while providing a simple scoring matrix.2 The standardization enables interpatient and intrapatient comparisons (e.g., before and after fitting of a hydraulic knee). However, most of the physical tasks performed are partially representative of the full range of ADL. Indeed, evidence that K-levels correlate with a 6-minute walk test, for example, are unsatisfactory.9 Furthermore, the predictive ability of these instruments of actual functional outcome is limited. Several studies demonstrated that amputees do not reliably self-report their ADL.10 For instance, comparison among 2-minute walk, timed get-up and go, locomotor capabilities index, and self-report performance measurement showed moderate correlations.4,11
PHYSICAL MEASUREMENTS OF ADL
Alternatively, the actual functional outcome, defined as “a participant's mobility with his or her prosthesis and ambulation aids in the home and community environment,”4 can be assessed after fitting of the prosthesis using physical measurements during real world ADL.
The most sophisticated pedometers (e.g., Step Activity Monitor) have the capacity to record continuously the number of steps and the cadence for periods of days to months.12–15 A recent study demonstrated that 2-minute walk test was highly correlated with step counts.4 Pedometers are accurate to detect distinct gait cycles. However, they provide an incomplete description of the level of intensity and type of activities, particularly the ones that are not derived from steps.
Other studies used a portable kinematic recording system based on several 2D or 3D accelerometer-type sensors (e.g., Patient Activity Monitor) to monitor the frequency and duration of activities in patients' habitual environments over several hours.16–18 Studies focusing on other populations (e.g., total hip replacement) completed analysis of the raw data by implementing an algorithm recognizing typical patterns of certain ADL, such as lying, sitting, standing, level and incline walking, and ascending and descending stairs.16 These analyses have the potential to give a more comprehensive and realistic insight into the actual functional outcome, provided that the population involved presents small variability of kinematic patterns. However, activities that are unclearly defined are dismissed although they might represent a significant portion of time.
NEED FOR COMPREHENSIVE ASSESSMENT OF ADL
All combined, these studies demonstrated a lack of correspondence within surrogate measurements and between surrogate and physical measurements. This might be because ecological measurements during actual ADL might not be sufficiently comprehensive and, more importantly, subjected to many more confounders (e.g., weather conditions, job demands, and marital status). Clearly, there is a need for an instrument capable of assessing the actual function outcome during ADL.
CATEGORIZATION OF THE LOAD REGIME DATA
More recently, a portable kinetic system, based on a transducer and data logger, was introduced for the continuous recording of the true load regime (i.e., frequency and magnitude of overall loading) applied on the residuum of a transfemoral amputee during ADL.19 This study presented only the recording of the raw data and some overall performance indicators of the usage of the prosthesis.
However, the opportunities to use these load regime data to assess the actual functional outcome during ADL are yet to be explored fully. This could be achieved using the following approaches:
- Recognition of activities: set activities could be recognized using templates of patterns that have been established after controlled measurements of individual standardized activities (e.g., descending stairs).16,20 These templates must be patient specific and, therefore, not easily transferable, given gait variability within a population of amputees.21–23 Also, templates are not always applicable to real world measurements because of variations in design of environment (e.g., height and depth of stairs) and amputee's ambulatory styles (e.g., descending front on or sideways, use of hand rail). This would be the conventional technique to assess the functional outcome, defined as the capacity to undertake recognizable, but limited, activities.
- Categorization of activities: as suggested by Frossard et al.,19 the raw load data could be split into categories of ADL, such as inactivity, stationary loading, and locomotion. The totality of the recording would be taken into consideration, instead of separate standard activities. Then, this innovative approach would assess the functional output, defined as the overall ability to undertake ADL.
In principle, the results of both techniques could be valuable for clinicians. However, the categorization of load regime data seems to be more straight forward (i.e., not preanalysis of standardized activities), complete (i.e., all activities included), and aligned with the underlying principles of Functional Classification Levels (i.e., determining overall ability).
PURPOSES AND OBJECTIVES
The purpose of this preliminary study was to determine the relevance of the categorization of the load regime data to assess the actual functional output and usage of the prosthesis of lower limb amputees. The objectives were a) to introduce a categorization of load regime, b) to present some descriptors of each activity, and c) to report the results for a case.
The raw load data used in this study have been published in the study by Frossard et al.19 along with the detailed account of methodological aspects including the portable kinetic system relying on a transducer and a data logger. Consequently, only the most relevant information is presented here.
One fully rehabilitated and active male (33 yr, 1.70 m, 85 kg, or 833.85 N, 12 yr since amputation) fitted with an osseointegrated fixation24–27 was asked to participate. He achieved an F in the SIGAM Scale (i.e., normal or near normal gait8), and therefore, he was classified as a K4 in the Functional Classification Levels (i.e., ability or potential for prosthetic ambulation that exceeds basic ambulation skills, exhibiting high impact, and stress or energy levels1). The research institution's human ethics committee approved this study. The participant provided informed written consent.
The prosthesis included a Rotasafe (Integrum AB, Göteborg, Sweden), a transducer, the participant's usual knee (Otto-Bock 3R80) and foot (Otto-Bock 1D10) fitted with hard running shoes as presented in Figure 2. The mass of the prosthesis below the transducer was approximately 0.65 kg. The forces and moments, commonly referred to as the load, were directly measured by a six-channel transducer (Model 45E15A; JR3 Inc., Woodland, CA), similar to the one used in a previous studies.19,28–30 The power was supplied by a customized battery pack placed in a waist pack attached to the subject. Data were processed using a calibration matrix to eliminate cross-talk between axial sensors. A preliminary experiment demonstrated that forces and moments along the three axes were measured by the transducer with an error of less than ±1 N and ±1 Nm.30 The transducer was mounted to customized plates that were positioned between the Rotasafe and the knee. These plates were used to anchor the transducer to pyramidal adaptors.
The data logger was connected to the transducer by a serial cable and placed in the waist pack. The output of the transducer was digitally stored using an 8-bit data logger (Valitec AD128, Daytona, OH) through additional interface circuitry. The 8-bit resolution of the data logger corresponds to a measurement resolution of approximately 8.95 N for the force along the long axis, 4.75 N for forces along the anteroposterior and mediolateral axes, 0.25 Nm for the moment about the long axis and 0.785 Nm for moments about the anteroposterior and mediolateral axes. The forces and moments were recorded with a sampling frequency of 10 Hz allowing a continuous monitoring period of 5 hr corresponding to 175,600 samples per channel, given the 2-Mb memory limitation of the data logger.
The prosthesis including the transducer was configured by a qualified prosthetist and fitted to the participant. The prosthetist attempted to align the leg as closely as possible to the usual alignment. The prosthetic leg was worn approximately 15 minutes before recording to ensure subject confidence and comfort.
The participant was asked to carry on his activities as normally as possible. The recording started shortly after 1:30 pm and lasted until 6:30 pm, giving a continuous recording of approximately 5 hours of the recreational afternoon, like comparable studies.16 The testing took place in January with an ambient temperature of approximately 17°C and overcast conditions, allowing the participant to carry on normal activities. He walked without aids.
Finally, the kinetic system was removed. It should be noted that the participant reported no problems wearing the apparatus, although the waist pack was found cumbersome in some instances (e.g., seating).
DATA PROCESSING: CATEGORIZATION
The load was divided into four categories of activities: directional locomotion, localized locomotion, stationary loading, and inactivity. An overview of the definition, the estimated range of displacement, the loading characteristics, and some typical examples of possible activities for each category is presented in Table 1. A combination of duration and magnitude of the signal were used to differentiate categories. These thresholds emanated from a heuristic approach and review of the literature focusing on classification and detection of ADL3,16,22–23 and preanalysis of the raw data. The typical examples of possible activities were presented to illustrate the type of activities that the participant might undertake. There were illustrative and tentative as no separate measurements (e.g., shadowing, pattern recognition) were conducted. For instance, the categorization encompassed a number of evaluation criteria of the Functional Classification Levels and other instruments. The periods of ambulation when the participant engaged into a displacement included directional and localized locomotion. The periods of activity included the periods of ambulation and the stationary loading. A customized Matlab software program (Math Works Inc, Natick, MA) was used to separate automatically each category. The detection of each activity consisted on recognizing first the inactivity, stationary loading, and directional locomotion activities, respectively. Activities that were not detected as one of these three activities were considered as localized locomotion. The program provided a reliable process including a faster computing time and consistent separation compared with manual technique. For instance, the software detected 98% of the phases picked manually for three random portions of the recording.
DATA ANALYSIS: CHARACTERIZATION
Each category was characterized by:
- General descriptors including the number of occurrences corresponding to the number of times an activity was detected and the duration of each category corresponding to the cumulated amount of time spent for each occurrence. They provided a broad insight on the functional level and usage of the prosthesis (e.g., activity vs. inactivity).
- Loading characteristics reflected by median, minimum, and maximum of the magnitude of the raw forces and moments applied on the three axes. Also, the duration of the resultant force between 12.5% and 37.5% and more than 50% of the body weight (BW) was assessed for each activity. All combined, these indicators reflected the loading abilities of the participant, which depend on comfort, confidence, relevant fitting, etc. For example, it is more likely that a load corresponding to 50% of BW during stationary loading reflects a well-fitted prosthesis.
- Impulse of the forces on the three axes, calculated using conventional trapeze methods based on the integration of the area under the force-time curves.31 This indicator summed up the overall usage of the prosthesis taking into consideration the magnitude and the duration of the load.19,21 The higher the value, the more the prosthesis was used. Gait cycles were subjected to complementary analysis. The temporal variables were extracted including the cadence expressed in number of strides of the prosthetic leg per minute, the duration of the gait cycle, and the support and swing phases expressed in seconds and percentage of gait cycle.
EXAMPLE OF RAW DATA
A sample of 2 minutes of recording presented in Figure 3 illustrated the identification of directional locomotion (e.g., 10 gait cycles, 0-15 seconds), localized locomotion (e.g., 15-30 seconds), stationary loading (e.g., 75-82 seconds), and inactivity (e.g., 90-110 seconds).
The occurrence and duration of each category of activities are presented in Table 2. The directional locomotion, localized locomotion, and stationary loading corresponded to 44%, 34%, and 22% of the occurrences and 51%, 38%, and 12% of the duration of the periods of activity, respectively. The ambulation represented more than 78% of the periods of activity.
The median, minimum, and maximum of the forces and moments applied along the three axes of the residuum for each category of activities are presented in Table 3. The absolute maximum force during directional locomotion, localized locomotion, and stationary loading represented 19%, 15%, and 8% of BW on the anteroposterior axis, 20%, 19%, and 12% on the mediolateral axis, and 121%, 106%, and 99% on the long axis, respectively. The minimum load applied on the long axis was negative (traction) because of gravity acting on the prosthetic components below the transducer during the swing phase. The resultant of the force was between 12.5% and 37.5% of the BW for 5%; 23% and 14%, and above half of the BW for 47%; 20% and 3% of the duration of directional locomotion, localized locomotion, and stationary loading, respectively. The minimum and maximum of the load registered during the inactivity category corresponded to odd movements produced to readjust the resting posture.
The impulse of the force for each category of activities is presented in Table 4. Approximately half of the total impulse of the resultant force was because of directional locomotion. Localized locomotion, stationary loading, and inactivity represented 31%, 11%, and 9%, respectively.
CHARACTERIZATION OF GAIT CYCLES
The participant generated a total of 2,783 gait cycles of the prosthesis during the recording period. Directional and localized locomotion included 90% (2,512) and 10% (271) of these gait cycles, respectively. The overall cadence was 10 strides/minute, but the more meaningful cadence during directional locomotion was 47 strides/minute. The mean duration of the gait cycle was 1.26 ± 0.16 seconds. The mean duration of the swing and support phases were 0.58 ± 0.12 seconds and 0.67 ± 0.09 seconds, corresponding to 46% and 54% of the gait cycle, respectively.
This preliminary study was designed to determine the feasibility of categorization of ADL alone, in contrast with a comparative study looking at the performance of the categorization in relation to other physical measurements or a case study discussing the participant's results.
Nonetheless, the recognition of activities is also partially validated. A study demonstrated that the direct measurements are as accurate as the ones obtained with inverse dynamics.30 Recognizing inactivity is rather straightforward. In principle, several studies focusing on load bearing exercises32 and activities of standardized22–23,28,33 and real world ADL19 could provide a surrogate validation of the recognition of the stationary loading and directional locomotion, respectively. By definition, localized locomotion is more an in-between activity difficult to validate. One way to validate the recognition would be to shadow the participant while tracking activity and filling a detailed diary. This is particularly challenging during ecological assessments (i.e., invasion of private space, getting in and out public or private transports) of rapidly changing activities.
Furthermore, the extraction of clinical information for this young and active participant is limited. The domination of long periods of inactivity might be explained by the fact that recording occurred during a recreational afternoon. More emphasis might have been placed on resting. However, a number of indicators demonstrated the ambulatory abilities of the participant and proper fitting of the prosthesis, including:
- The number of occurrences and duration of periods of activity and ambulation,
- The maximum loading on the long axis during directional locomotion,
- The duration of loading above and below half of the BW,
- The temporal characteristics of the gait cycles were in the upper end of the ones reported for transfemoral amputees in the previous studies.10,34–36
All combined the results concurred with previous self-report assessments (i.e., F in SIGAM and K4).
RELEVANCE OF PROPOSED CATEGORIZATION
This work highlighted the difficulty of achieving appropriate assessment of the true functional output and usage of the prosthesis with typical resources available in clinical settings.
This study demonstrated that the proposed categorization of ADL has the potential to provide a more comprehensive assessment than current instruments mainly because the measurements were not limited to directional locomotion. In this case, this enabled the detection of approximately 10% more gait cycles that were unlikely to be registered by conventional pedometers. Furthermore, it enabled the measurement of approximately 50% more of the total impulse, occurring during localized locomotion, stationary loading, and inactivity, that would have been difficult to estimate using conventional analysis (i.e., the number of steps measured by pedometers in real world13 multiplied by the impulse obtained in a gait laboratory for a few steps37,38).
Some of the conventional instruments require little resources compared with the proposed apparatus. Consequently, its systematic implementation in clinical settings is somewhat unrealistic. Nonetheless, one can argue that this type of assessment will be best used as a complement rather than a replacement of conventional instruments. For example, it will be relevant to differentiate difficult patients who are in-between K3 and K4 levels.
DEVELOPMENT OF FUTURE PROTOTYPES
From an engineering point of view, this study corresponded to a proof-of-concept study. Indeed, it provided sufficient technical information to further develop a fully functioning prototype of an instrument (i.e., hardware and software) specifically designed for clinical applications. This study revealed that a more compact recording device will be needed to reduce encumbrance (e.g., carrying batteries in waist pack). A recording frequency of up to 120 Hz will give a better insight into the maximum loading and impulse generated and more accurate detection of gait events such heelstrike transient, provided that it occurs close to 60 Hz. Like any other battery operated device, a recording capacity will reduce the number of shutdown times to change batteries and download data logger and, therefore, enabling more frequent recordings and representative snapshots of ADL. Finally, improving the clinician-software interface will be required to ease setting up of detection parameters and report of results. All these features could be easily implemented using a handheld computer, for example.
TOOL FOR CLINICAL STUDIES
The instrument presented here will facilitate longitudinal studies of ADL for a larger cohort of participants. This study involved an amputee fitted with an osseointegrated fixation because the initial purpose of the recording was to monitor the load regime applied on the residuum to better design the fixation (e.g., fatigue and fracture).19 However, it is important to indicate that a similar analysis could have been conducted on any other lower limb amputees fitted with a socket (e.g., quadrilateral,28,30 ischial containment). The proposed categorization can be done regardless of the attachment, providing that the transducer can be mounted within the prosthesis above or below the knee (e.g., pylon). Such longitudinal studies will provide a better understanding of the participant-to-participant variability on level and category of activities. Some confirmation of the K level and SIGAM score were provided. However, a comprehensive comparison of the results from other instruments (i.e., self-report and physical tasks) and the proposed categorization was outside the scope to this proof-of-concept study. However, the possibilities for cross-sectional studies are endless, particularly for the ones allowing reciprocal validation of these instruments (e.g., number of steps measured with the transducer and Step Activity Monitor) and correlation of the outcomes (e.g., K-level and impulse during directional locomotion). Further work is needed to identify the activities undertook in each category.
Both longitudinal and cross-sectional studies will be essential to improve basic knowledge in the areas of rehabilitation (e.g., loading technique and usage of walking aids29), design of components (e.g., fatigue, loading requirement, product classification, and fall detection), and fitting of prosthesis (e.g., alignment, threshold of protective device, and prescription of components). They might also help to refine the definition of activity and function outcome and standard of activity levels as presented in the World Health Organization International Classification of Functioning, Disability and Health.39
A categorization of ADL based on a portable kinetic system has been presented that enables the characterization of the actual functional output and usage of the prosthesis. An example of raw results and some of the derived information were provided for one transfemoral amputee to illustrate the capacities of this new categorization.
This study highlighted some shortcomings of the current instruments measuring physical variables in real world settings. This study established that the core principle underlying categorization of activities have the potential to provide more comprehensive outcomes than the recognition of activities, because it takes into consideration activities other than directional locomotion.
In conclusion, the categorization presented here is a stepping stone in the development of a user-friendly instrument based on a portable kinetic system to be used by clinicians responsible for outcome measures and classification of lower limb amputees.
The authors wish to acknowledge Prof. John Evans, Rickard Brånemark, Eva Häggström, Kerstin Hagberg, Chris Daniel, Kingsley Robinson, David Lee Gow, and Heather Curtis for their valuable contribution to the data collection.
1. Gailey RS, Roach KE, Applegate EB, et al. The Amputee Mobility Predictor: an instrument to assess determinants of the lower-limb amputee's ability to ambulate. Arch Phys Med Rehabil
2. Miller L, McCay JA. Summary and conclusions from the academy's sixth state-of-the-science conference on lower limb prosthetic outcome measures. J Prosthet Orthot
3. Condie E, Scott H, Treweek S. Lower limb prosthetic outcome measures: a review of the literature 1995 to 2005. J Prosthet Orthot
4. Parker K, Kirby RL, Adderson J, Thompson K. Ambulation of people with lower-limb amputations: relationship between capacity and performance measures. Arch Phys Med Rehabil
5. Gauthier-Gagnon C, Grisé M-C. Tools to measure outcome of people with a lower limb amputation: update on the PPA and LCI. J Prosthet Orthot
6. Boone DA, Coleman KL. Use of the Prosthesis Evaluation Questionnaire (PEQ). J Prosthet Orthot
7. Hagberg K, Brånemark R, Hagg O. Questionnaire for Persons with a Transfemoral Amputation (Q-TFA): initial validity and reliability of a new outcome measure. J Rehabil Res Dev
8. Ryall N, Eyres S, Neumann V, et al. The SIGAM mobility grades: a new population-specific measure for lower limb amputees. Disabil Rehabil
9. Gailey RS. Predictive outcome measures versus functional outcome measures in the lower limb amputee. J Prosthet Orthot
10. Stepien JM, Cavenett S, Taylor L, Crotty M. Activity levels among lower-limb amputees: self-report versus Step Activity Monitor. Arch Phys Med Rehabil
11. Miller WC, Deathe AB, Speechley M. Lower extremity prosthetic mobility: a comparison of 3 self-report scales. Arch Phys Med Rehabil
12. Day HJB. The assessment and description of amputee activity. Prosthet Orthot Int
13. Boone DA, Coleman KL. Use of a Step Activity Monitor in determining outcomes. J Prosthet Orthot
14. Ramstrand N, Nilsson K-Ã. Validation of a patient activity monitor to quantify ambulatory activity in an amputee population. Prosthet Orthot Int
15. Coleman KL, Smith DG, Boone DA, et al. Step activity monitor: long-term continuous recording of ambulatory function. J Rehabil Res Dev
16. Morlock M, Schneider E, Bluhm A, et al. Duration and frequency of every day activities in total hip patients. J Biomech
17. Tryon WW. Measuring activity using actometers: a methodological study. J Psychopathol Behav Assess
18. Jafari R, Li W, Bajcsy R, et al. Physical activity monitoring for assisted living at home. In: Leonhard S, ed. 4th International Workshop on Wearable and Implantable Body Sensor Networks
; Aachen, Germany: Springer; 2007:213–219.
19. Frossard L, Stevenson N, Smeathers J, et al. Monitoring of the load regime applied on the osseointegrated fixation of a transfemoral amputee: a tool for evidence-based practice. Prosthet Orthot Int
20. Oshima Y, Kawaguchi K, Tanaka S, et al. Classifying household and locomotive activities using a triaxial accelerometer. Gait Posture
21. Zahedi MS, Spence WD, Solomonidis SE, Paul JP. Repeatability of kinetic measurements in gait studies of the lower limb amputee. Prosthet Orthot Int
22. Lee W, Frossard L, Hagberg K, et al. Kinetics analysis of transfemoral amputees fitted with osseointegrated fixation performing common activities of daily living
. Clin Biomech
23. Lee WC, Frossard LA, Hagberg K, et al. Magnitude and variability of loading on the osseointegrated implant of transfemoral amputees during walking. Med Eng Phys
24. Brånemark R, Brånemark P-I, Rydevik B, Myers R. Osseointegration in skeletal reconstruction and rehabilitation: a review. J Rehabil Res Dev
25. Sullivan J, Uden M, Robinson K, Sooriakumaran S. Rehabilitation of the trans-femoral amputee with an osseointegrated prosthesis: the United Kingdom experience. Prosthet Orthot Int
26. Hagberg K. Chapter 25: Physiotherapy for patients having a trans-femoral amputation. In: Brånemark P-I, editor. The Osseointegration Book—from Calvarium to Calcaneus
. Berlin: Quintessenz Verlag-GmbH; 2005:477–487.
27. Pitkin M. On the way to total integration of prosthetic pylon with residuum. J Rehabil Res Dev
28. Frossard L, Beck J, Dillon M, et al. Development and preliminary testing of a device for the direct measurement of forces and moments in the prosthetic limb of transfemoral amputees during activities of daily living
. J Prosthet Orthot
29. Frossard L, Hagberg K, Haggstrom E, Branemark R. Load-relief of walking aids on osseointegrated fixation: instrument for evidence-based practice. IEEE Trans Neural Syst Rehabil Eng
30. Dumas R, Cheze L, Frossard L. Loading applied on prosthetic knee of transfemoral amputee: comparison of inverse dynamics and direct measurements. Gait Posture
31. Seliktar R, Yekutiel M, Bar A. Gait consistency test based on the impulse-momentum theorem. Prosthet Orthot Int
32. Frossard L, Gow DL, Hagberg K, et al. Apparatus for monitoring load bearing rehabilitation exercises of a transfemoral amputee fitted with an osseointegrated fixation: a proof-of-concept study. Gait Posture
33. Frossard LP, Hagberg KP, Haggstrom ECPO, et al. Functional outcome of transfemoral amputees fitted with an osseointegrated fixation: temporal gait characteristics. J Prosthet Orthot
34. Skinner H, Efferney D. Special review “Gait analysis in amputees.” Am J Phys Med
35. Murray MP, Mollinger LA, Sepic SB, Gardner GM. Gait patterns in above-knee amputee patients: hydraulic swing control vs constant-friction knee components. Arch Phys Med Rehabil
36. Murray MP, Sepic SB, Gardener G, Mollinger L. Gait patterns of above-knee amputees using constant friction knee components. Bull Prosthet Res
37. DiAngelo DJ, Winter DA, Ghista DN, Newcombe WR. Performance assessment of the Terry Fox jogging prosthesis for above-knee amputees. J Biomech
38. Stephenson P, Seedhom BB. Estimation of forces at the interface between an artificial limb and an implant directly fixed into the femur in above-knee amputees. J Orthop Sci
39. Stucki G, Ewert T, Cieza A. Value and application of the ICF in rehabilitation medicine. Disabil Rehabil